When Accuracy Is Not Enough: Uncertainty Collapse between Noisy Label Learning and Out-of-Distribution Detection
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866914576258301952 |
|---|---|
| author | Peng, Ningkang Mao, Jingyang Zhou, Runhan Ma, Peirong Gu, Yanhui |
| author_facet | Peng, Ningkang Mao, Jingyang Zhou, Runhan Ma, Peirong Gu, Yanhui |
| contents | Learning with noisy labels (LNL) is typically benchmarked by closed-set classification accuracy, yet deployment often requires classifiers to reject out-of-distribution (OOD) inputs. We present a learner-agnostic ACC-OOD benchmark that freezes LNL checkpoints and evaluates them with standardized near-/far-OOD routing and post-hoc scores across synthetic and real label noise. The benchmark reveals a recurring failure mode: high closed-set accuracy does not ensure OOD reliability, because low-confidence, misclassified in-distribution samples can overlap the score and feature regions occupied by OOD inputs under noisy training. We term this pathology uncertainty collapse. This structural overlap can make high-accuracy LNL methods lose separability at the ID-error/OOD interface under standard OOD scores. As an intervention, we study Virtual Margin Regularization (VMR), a lightweight repair probe demonstrated mainly with PSSCL that synthesizes boundary virtual outliers on trusted ID batches and widens the energy margin. VMR partially reduces the collapse-induced far-OOD failure without replacing the host objective or sacrificing closed-set accuracy in the tested settings. These results support LNL benchmarks that co-report closed-set generalization, open-world reliability, and structural overlap diagnostics. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_17795 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | When Accuracy Is Not Enough: Uncertainty Collapse between Noisy Label Learning and Out-of-Distribution Detection Peng, Ningkang Mao, Jingyang Zhou, Runhan Ma, Peirong Gu, Yanhui Machine Learning Computer Vision and Pattern Recognition Learning with noisy labels (LNL) is typically benchmarked by closed-set classification accuracy, yet deployment often requires classifiers to reject out-of-distribution (OOD) inputs. We present a learner-agnostic ACC-OOD benchmark that freezes LNL checkpoints and evaluates them with standardized near-/far-OOD routing and post-hoc scores across synthetic and real label noise. The benchmark reveals a recurring failure mode: high closed-set accuracy does not ensure OOD reliability, because low-confidence, misclassified in-distribution samples can overlap the score and feature regions occupied by OOD inputs under noisy training. We term this pathology uncertainty collapse. This structural overlap can make high-accuracy LNL methods lose separability at the ID-error/OOD interface under standard OOD scores. As an intervention, we study Virtual Margin Regularization (VMR), a lightweight repair probe demonstrated mainly with PSSCL that synthesizes boundary virtual outliers on trusted ID batches and widens the energy margin. VMR partially reduces the collapse-induced far-OOD failure without replacing the host objective or sacrificing closed-set accuracy in the tested settings. These results support LNL benchmarks that co-report closed-set generalization, open-world reliability, and structural overlap diagnostics. |
| title | When Accuracy Is Not Enough: Uncertainty Collapse between Noisy Label Learning and Out-of-Distribution Detection |
| topic | Machine Learning Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2605.17795 |